We showed that eukaryotic replication termination occurs at TERs containing fork barriers. There are intriguing analogies with prokaryotes where specific termination sites and polar pausing elements influence termination. It is possible that fork barriers have passively localized through evolution in proximity of TERs, because if replication forks have to pause, it is least disadvantageous when this occurs at a site where forks are converging. Alternatively, evolution has brought fork barriers at TERs to influence fork fusion. Intriguingly, we note that deleting an efficient origin causes the re-localization of fork fusion from the original TER to another pausing element (data not shown), thus suggesting that the site of termination is influenced by the presence of pause sites.
Our findings also suggest that the polarity of fork barriers had an evolutionary impact on chromosome replication and on TERs
integrity. Indeed using the yeast comparative genomics database we notice that in 5/6 TERs
(TER304, 702, 801, 1601
) containing two divergent Pol III-dependent pause sites (tRNA/LTR), one of them is totally or partially not conserved (Figure S5
and Ted Weinert personal communication). On the other hand, those 58 TERs
that contain polar barriers have conserved the pause sites in other yeasts. We excluded from the analysis the 7 TERs
-containing centromeres as CENs
are known to rapidly diverge in evolution (Henikoff et al., 2001
) (and on the other side represent bipolar pausing elements). This correlation (p= 0.00000465) further suggests the existence of an evolutionary pressure against TER
-containing pause sites on both strands perhaps to avoid genome instability events. In this view, we note that TER502
(the remaining un-conserved TER
are unstable in top2
mutants as shown by CGH analysis (), TER304
are hot spots for genome rearrangements (Admire et al., 2006
; Lemoine et al., 2005
), and γH2A accumulates in TER304, 502, 702
). It will be of interest to address how replication termination is achieved when transcription is dispensable as in the frog embryonic cell cycle. We also note that TERs
seem to correlate with low nucleosome regions (p=0,07) (Table S5
A model for replication termination
- Rrm3, Top1 and a fraction of Top2 travel with the fork (Azvolinsky et al., 2006; Bermejo et al., 2007). Rrm3 facilitates forks progression across pausing sites (Ivessa et al., 2003) while Top1 and Top2 are both needed to resolve the torsional stress ahead of the fork generated during fork progression: while Top1 resolves positive supercoiling ahead of the fork (Wang, 2002), also contributing to prevent interference between replication and transcription (Tuduri et al., 2009), Top2 likely acts behind the fork to resolve precatenanes (Lucas et al., 2001; Wang, 2002). When forks approach the termination zone, the topological constrains at converging forks can no longer be resolved by Top1 (Fields-Berry and DePamphilis, 1989) and therefore the only option for fork progression is to rely on Top2 activity. This is consistent with the observation that top2 mutants are selectively delayed in completing the last portion of replication but not the bulk of DNA synthesis. However, we cannot rule out that the topological architecture of the termination zone (e.g. chromosome loops) needs specifically Top2 activity for resolution. Indeed a subpopulation of Top2 is also bound to TERs in early S-phase, perhaps due to the affinity of Top2 for nucleosome-free regions (p= 2.10E-58). Moreover, other S-phase Top2 clusters have been recently suggested to correlate with the formation of chromosome loops (Bermejo et al., 2009). We found that the Top2 clusters at TERs are established already at the cdc7 dependent step and are not influenced by origin firing (data not shown), thus suggesting that TERs represent CIS chromosomal elements that undergo topological transitions requiring Top2 activity.
- When fork fusion occurs, the lagging polymerase encounters the leading strand polymerase from the opposite fork, thus physically occupying the remaining un-replicated region (Sundin and Varshavsky, 1981). It is still unclear how the replisome is dismantled and how fork fusion occurs. Perhaps the presence of polar fork barriers may guarantee that the two forks do not converge simultaneously thus ensuring that at least one of the two forks emerges from the pausing region with asymmetric leading and lagging strands before fusing with the other fork. This is consistent with the findings that stalled forks exhibit an asymmetric configuration (Gruber et al., 2000; Sogo et al., 2002). Rrm3 could simply facilitate fork progression at the pause sites located within the TERs. However, we cannot exclude the possibility that Rrm3 actively participates at fork fusion as suggested by the finding that unresolved termination structures accumulate even at those TERs that do not contain obvious Rrm3-dependent pause elements.
Considering that i) the termination context might be ideal for fork reversal as topological constrains accumulate and the replisome must be dismantled (Postow et al., 2001
) ii) the Mec1-Rad53 checkpoint pathway prevents fork reversal when forks stall (Sogo et al., 2002
) iii) checkpoint factors have been implicated in mediating termination at the rDNA locus (Mohanty et al., 2006
), it is tantalizing to speculate that the Mec1 checkpoint pathway somewhat prevents aberrant fork transitions, such as fork reversal, at termination zones.
- Fork fusion then gives rise to catenated sister chromatid junctions that have to be resolved before segregation. We propose that this last step is mediated by a sub-population of pre-assembled TER-associated Top2 that can persist even after S-phase. It is also possible that Top2, at least in a fraction of TERs, is loaded at the beginning of mitosis. Given that the catenated junction might be mobile and spread along the chromosomes (Spell and Holm, 1994), the presence of preassembled Top2 might be needed to confine and coordinate its resolution at the TER loci, perhaps through SUMO-mediated regulation (Bachant et al., 2002; Dawlaty et al., 2008).
According to the model proposed, the transient accumulation of topological constrains might facilitate abnormal transitions (Hiasa and Marians, 1994
) that could lead to amplification or deletion of TER
sites. Moreover, the proper resolution of catenated sister chromatids would be impaired in top2
cells and, following cell division, DNA breaks, and aberrant segregation will be expected (Baxter and Diffley, 2008
; Bermejo et al., 2007
; DiNardo et al., 1984
; Holm et al., 1989
Altogether our data provide a framework for understanding the eukaryotic molecular mechanisms that control replication termination and coordinate replication with transcription and topological dynamics.